In long distance optical communication systems it may be important to monitor the health of the system. For example, monitoring can be used to detect faults or breaks in the optical transmission cable, faulty repeaters or amplifiers or other problems with the system.
Known monitoring techniques include use of line monitoring equipment that generates a test signal representing a pseudo random bit sequence. The line monitoring equipment may transmit the test signal with the information signals, e.g. in wavelength division multiplexed system. The test signal may be returned to the line monitoring equipment through a high-loss loopback (HLLB) path within an amplifier or repeater. The line monitoring equipment may then separate the returned test signal from the data signals, and process the returned test signal to obtain data representing the HLLB loop gain imparted to the test signal in its propagation from the line monitoring equipment, through the HLLB and any intervening optical paths and amplifiers, and back to the line monitoring equipment. Significant deviations in HLLB loop gain may indicate a fault in the system.
In an undersea optical communication system, repeater pump power loss and increased fiber span loss may be primary failure mechanisms resulting in HLLB loop gain deviations from normal values. In a known system, significant variations in HLLB loop gain, e.g. above a predefined alarm threshold, may trigger a system alarm. Choice of the alarm threshold in such a system may require discrimination between normal system fluctuations and measurement errors and real transmission path faults. Unfortunately, this discrimination may be difficult since HLLB loop gain measurements may be generally insensitive to physical changes in the transmission path due, in part, to the repeater loop back output-to-output architecture, as well as gain mechanisms in the repeater amplifier, e.g. self-gain regulation Consequently, real path changes for non-devastating failures in such changes may result in HLLB loop gain changes that are only slightly detectable given typical measurement errors and system fluctuations.